magico a3 vs. revel 228 be
TRANSCRIPT
HARMAN International. Confidential. Copyright 2017.
MAGICO A3 VS. REVEL 228 BE
SEAN OLIVE, TODD WELTI, OMID KHONSARIPOUR
DECEMBER 10 2018
©2016 HARMAN INTERNATIONAL INDUSTRIES, INCORPORATED !2
• 9 listeners participated (all Harman employees)
• Tests performed double blind in Harman Multi-channel Lab
• Total of 4 trials: 2 tracks x 2 observations
• Speakers level-matched within 0.1 dB; playback level fixed at 80 dB (B-weighted) average
• Speaker and Program order randomized
• Speaker Position controlled via automated speaker mover
TEST METHOD
©2016 HARMAN INTERNATIONAL INDUSTRIES, INCORPORATED !3
• A repeated measures ANOVA shows that Loudspeaker was the dominant significant effect on listener preference • Program was not significant • There was a very small but not statistically significant interaction between Loudspeaker * Program
ANOVA
Source DF Sum of squares Mean squares F Pr > F Adj. Pr>F G-G Adj. Pr>F H-F
Loudspeaker 2 239.771 119.886 74.364 < 0.0001 < 0.0001 < 0.0001
Program 3 1.765 0.588 0.527 0.667
Program *Loudspeaker 6 20.468 3.411 2.116 0.058 0.061 0.058
Error 64 103.177 1.612
©2016 HARMAN INTERNATIONAL INDUSTRIES, INCORPORATED !4
• Box plot shows the distribution of ratings including mean (+), median, upper and lower quartiles and range of ratings
• Both Revel models were strongly preferred (> 2 preference ratings) to Magico A3
• Revel Salon 2 was moderately preferred (> 1 preference rating) to Revel 228 Be
• The differences in preference ratings among all loudspeakers were statistically significant based on Tukey HD post-hoc test at a level of 5%
MEAN SPEAKER PREFERENCES
MagicoA3
Revel228Be
RevelSalon-2
0
2
4
6
8
10
12
Preferen
ceRating
Box plots
©2016 HARMAN INTERNATIONAL INDUSTRIES, INCORPORATED !5
• All listeners rated Magico as their least favorite choice
• 7 of 9 listeners on average rated Revel Salon 2 higher than Revel 228 BE. One listener tied it with Revel 228 BE
• One listener (listener 8) preferred Revel 228 BE to Salon
INDIVIDUAL LISTENER PREFERENCES
Pref
eren
ce R
atin
g0
2
4
6
8
10
12
Listener
1 2 3 4 5 6 7 8 9
Magico A3 Revel 228 BE Revel Salon 2
©2016 HARMAN INTERNATIONAL INDUSTRIES, INCORPORATED !6
ALTERNATIVE VS. STANDARD TEST TRACKS
Pref
eren
ce R
atin
g
2
34
57
89
10
ProgramsDiana Krall Depeche Mode Jennifer Warnes Tracy Chapman
Magico A3 Revel 228 BE Revel Salon 2
Alternative Tracks Standard Tracks
Two listeners requested alternative test tracks (DK and DM) because they were more familiar with them and felt they could better hear and rate the differences among the speakers more accurately and reliably
However, the results suggest the opposite effect occurred:
The alternative tracks (DK and MD) produced less discriminating and consistent ratings compared to the standard tracks with DK being the worse case.
The standard tracks (JW and TC) produced higher ratings for Revel Speakers and lower ratings for Magico.
©2016 HARMAN INTERNATIONAL INDUSTRIES, INCORPORATED
The three speakers were measured in the same chamber by Mark Glazer using different measurement systems. There are errors below 100 Hz due to chamber calibration errors. Still, there are some clear differences in performance among the speakers:
1. Revel has flatter on-axis response with smoother off-axis, sound power curves and DI indices 2. Magico has some notable resonances at 150 Hz and 600 Hz. The 5-6 dB downward tilt from 80 Hz to 4 kHz makes
the loudspeaker sound imbalanced as noted in listener comments: dark, tubby, muffled, colored mids.
Magico A3 Revel Performa 228 BE
ANECHOIC MEASUREMENTSRevel Salon 2